safety

Optimizer-dependent training dynamics converge to the same one-third optimal data scaling

arXiv:2609.37745v1 Announce Type: new Abstract: Neural scaling, in which loss falls as a power law with training, is central to large language models, and one recent proposal is that a 1/3 exponent em

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

OTROPE: Optimal Transport-based Robust Off-policy Evaluation for Large Language Models

DGX agent

arXiv:2609.36264v1 Announce Type: new Abstract: Reliable evaluation of large language models (LLMs) is essential for their development and deployment, yet is often costly, risky, and difficult to perf

safetyarxiv-cs-ai

30 Sept 2026

Overcoming Scaling Limits in On-Policy Self-Distillation for LLM Reasoning

arXiv:2609.37915v1 Announce Type: cross Abstract: On-policy self-distillation (OPSD) trains a student to match a privileged teacher distribution along its own sampled trajectory. Standard OPSD applies

PADME: Preference Alignment Data Synthesis for Meta-Evaluation of LM Agent Evaluators

arXiv:2609.36086v1 Announce Type: cross Abstract: Language models are frequently employed to evaluate other language models. An LM evaluator scoring agentic behaviors across multiple criteria is valua

Pair Difficulty Matters: Rethinking Pairwise LLM-as-a-Judge Evaluation and Consistency

arXiv:2609.37577v1 Announce Type: new Abstract: Large Language Model judges are widely used to rank texts and text-generating systems through pairwise comparison, and their reliability is typically as

PE-OPSD: Internalizing Prompt Enhancement into Flow-matching Models via On-Policy Self-Distillation

arXiv:2609.36638v1 Announce Type: cross Abstract: Text-to-image users often provide concise and underspecified prompts, whereas generative models benefit from detailed textual conditions for reliable

Physics-Aware Machine Unlearning for Cyber-Physical Systems

arXiv:2609.36633v1 Announce Type: new Abstract: This paper proposes a physics-guided gradient-ascent-based machine unlearning method that couples the forgetting signal with the physical residual of th

Physics-Informed Multi-Agent Coordination for Hospital Patient Flow Optimization

arXiv:2609.37022v1 Announce Type: new Abstract: Efficient patient flow coordination across autonomous hospital departments is critical for mitigating overcrowding and balancing resource utilization. W

POET: Preference Optimization for Enhanced Text-to-Image Generation

arXiv:2510.12041v3 Announce Type: replace Abstract: Recent advances in text-to-image (T2I) generation have achieved impressive results, yet existing models often struggle with simple or underspecified

Pointwise or Pairwise: When Do Pairwise Losses Help Reward Learning, Provably?

arXiv:2609.37209v1 Announce Type: new Abstract: Pairwise losses are increasingly used for reward learning even when pointwise rewards are observed, with mixed empirical results. When and why do pairwi

Population Fidelity: Evaluating Population Representativeness in LLMs

arXiv:2609.36253v1 Announce Type: cross Abstract: Large language models (LLMs) show considerable potential in simulating human attitudes and preferences. Prior work finds that LLM-generated responses

PowerSim: Differentiable Physics Simulation and Rendering with Power Diagrams

arXiv:2609.38153v1 Announce Type: new Abstract: We introduce PowerSim, a method to bring physically grounded, differentiable dynamics to PowerFoam's power diagram based 3D representation. PowerSim dir

PR-OPD: Privileged Representation On-policy Self-Distillation for Agentic Reinforcement Learning

arXiv:2609.36642v1 Announce Type: new Abstract: Language-model agents are usually trained by reinforcement learning from one reward per episode, and privileged self-distillation enriches it by letting

Predictive Safety Curricula for Robust Legged Locomotion

arXiv:2609.37070v1 Announce Type: cross Abstract: Rare but consequential failures can persist in learned locomotion policies for legged robots even when average task performance is high, in part becau

Preferent Compression Bounds Are Tight

arXiv:2609.36030v1 Announce Type: new Abstract: The lack of rigorous safety and performance certificates remains a key bottleneck to the deployment of modern learning-based methods. Sample compression

PrimeSeeker: Capability-Oriented Supervision for Deep Search Agents

arXiv:2609.35816v1 Announce Type: cross Abstract: Large language model search agents are often trained with synthetic questions whose difficulty is increased through larger evidence graphs, additional

Privy to the Foil: Recasting Value Estimation with a Self-Privileged Critic for RLVR

arXiv:2609.37825v1 Announce Type: cross Abstract: Assigning credit to intermediate steps remains a central challenge in training Large Language Models (LLMs) on multi-step reasoning tasks with sparse

Prominent tech CEOs sign voluntary White House AI safety accord

The chief executives of five tech giants have signed a voluntary artificial intelligence safety accord backed by the White House. U.S. President Donald Trump posted the document on Truth Social late T

safetysiliconangle

Prompts Live on an Arc: Gaussian Curricula in Fisher--Rao Coordinates for Rollout-Efficient GRPO

arXiv:2609.38018v1 Announce Type: new Abstract: Group relative policy optimization (GRPO) learns only from prompts whose sampled responses disagree: a group that is entirely correct or entirely incorr

Pruning for Efficiency, Paying in Fairness: Demographic Disparities in Pruned Speech-LLMs

arXiv:2609.38106v1 Announce Type: cross Abstract: Speech-LLMs are expensive to run, making compression important for real-world deployment. However, compressed models are usually selected using aggreg

Question-Specific Knowledge Graphs for Efficient Visual Reasoning

arXiv:2609.35942v1 Announce Type: new Abstract: Recent work in visual question answering has shown that vision-language models can exhibit strong reasoning capabilities by translating visual inputs in

RA-CFGCache: From Branch-Level Criteria to Guided-Risk Control under Classifier-Free Guidance

arXiv:2609.36433v1 Announce Type: new Abstract: Diffusion models enable high-quality visual generation, but iterative denoising remains computationally expensive, especially under classifier-free guid

RankBuffer: Efficient Ranking-Based Rewards for Open-Ended Generation

arXiv:2609.36652v1 Announce Type: new Abstract: Open-ended generation lacks canonical answers, making pointwise rewards difficult to calibrate for group-based reinforcement learning. Directly ranking

Rational Clarification by Assistive Agents via Value-of-Information Reasoning

arXiv:2609.37588v1 Announce Type: new Abstract: Users of language-based assistive agents often make ambiguous requests. In response, an assistant can either directly act on its interpretation of the r

RBF-GNN: Rational Basis Functions for Pseudo-Coordinate based Graph Convolutions

arXiv:2609.37015v1 Announce Type: new Abstract: We propose RBF-GNN, a new pseudo-coordinate based graph neural network architecture that takes into account Euclidean, spherical or angular coordinates

Reactive Real-Time Flow Policies via Asynchronous Distribution Alignment

arXiv:2609.36540v1 Announce Type: new Abstract: Generalist robot policies such as vision-language-action models (VLAs) have achieved remarkable generalization, but their inference delays can conflict

Real2Gym: Building Gyms from Videos, Bringing Skills to Robots

arXiv:2609.37089v1 Announce Type: new Abstract: Real-world videos provide rich demonstrations of manipulation, but turning them into reusable robot skills requires visually aligned environments, execu

ReF-HIL: Shaping the Critic around Human Action Neighborhoods for Efficient Human-in-the-Loop Reinforcement Learning

arXiv:2609.37131v1 Announce Type: new Abstract: Human-in-the-loop reinforcement learning (HIL-RL) offers a promising route to efficient training of robotic manipulation policies by combining autonomou

Regime Boundary Alignment for Evidence-Gated Question Answering

arXiv:2609.37491v1 Announce Type: cross Abstract: Retrieval-augmented language models are expected to answer from the retrieved evidence, but in practice they often keep answering when that evidence i

Remember What You Did: Action-History Memory with Dual-Expert Denoising for Long-Horizon Vision-Language-Action Policies

arXiv:2609.37307v1 Announce Type: new Abstract: Vision-language-action (VLA) models have driven rapid progress in robotic manipulation, demonstrating strong fine-grained control and promising performa

Representation by Design in Generation: Cross-View Class-Token Alignment in Diffusion Transformers

arXiv:2609.36348v1 Announce Type: cross Abstract: Generative and representation learning remain asymmetrically connected: semantic representations are used to improve diffusion generation, whereas the

Rethinking Representations for World-Action Modeling

arXiv:2609.38163v1 Announce Type: new Abstract: World-action models jointly learn robot policies and predict future observations, making the representation space an interface between control and predi

Retrieval Sensitivity to Identity Signals in Queries

arXiv:2609.36534v1 Announce Type: new Abstract: Dense retrievers decide which documents reach users and the language models that use them, yet they are typically evaluated with neutral queries. We ask

REVO: Rollout-Efficient Off-Policy Distillation via Variance-Guided Reuse

arXiv:2609.37500v1 Announce Type: new Abstract: On-policy distillation (OPD) trains language models using dense token-level teacher supervision on student-generated trajectories. However, its reliance

Reward-rate Policy Gradient for Efficient Machine Learning Engineering Agents

arXiv:2609.36393v1 Announce Type: cross Abstract: Traditional reinforcement learning (RL) techniques focus on maximizing expected cumulative reward, where each action assumes to take a constant unit o

Rho: A Foundation for Efficiently Adaptable VLA Models

arXiv:2609.38164v1 Announce Type: new Abstract: General-purpose physical AI models must combine broad visual and linguistic capabilities with precise control across robot embodiments and efficient ada

Right Words, Wrong Moment: A Clinician-Grounded Analysis of Distress in 19,930 Conversations between Young People and ChatGPT

arXiv:2609.35953v1 Announce Type: new Abstract: Young people increasingly turn to General-Purpose Conversational Agents (GPCAs), such as ChatGPT, in moments of distress. We examine young adults' (ages

Risk-Averse Online POMDP Planning via CVaR of the Immediate Cost with Performance Guarantees

arXiv:2609.35874v1 Announce Type: new Abstract: Online POMDP planners optimize the expected cumulative cost, which can mask dangerous states when the belief places significant mass on high-cost states

RL-PaO: Prediction as Action in Decision Making under Uncertainty

arXiv:2609.37065v1 Announce Type: new Abstract: Decision-making under uncertainty often relies on predicted parameters, yet accurate prediction does not necessarily lead to good operational decisions.

RNA Design via Conditioned Flow Matching and Finite-Policy Reinforcement Learning

arXiv:2609.36885v1 Announce Type: cross Abstract: RNA design aims to identify sequences that fold into specified secondary structures. Existing methods formulate the task as target-specific search or

RoXDrive: Closed-Loop Reinforcement Learning for End-to-End Autonomous Driving via Action-Faithful Rollouts

arXiv:2609.36851v1 Announce Type: new Abstract: End-to-end autonomous driving policies are commonly trained via imitation learning on logged demonstrations without observing the consequences of their

RS-OPSD: Reliable Privileged On-Policy-Self-Distillation for Ultra-High-Resolution Remote Sensing VQA

arXiv:2609.38072v1 Announce Type: new Abstract: Ultra-high-resolution (UHR) remote sensing visual question answering (VQA) requires models to resolve small visual evidence within extremely large image

Safe-by-Design Learning via Energy-based Neural Networks

arXiv:2609.36942v1 Announce Type: cross Abstract: Learning neural-network models of dynamical systems with safety guarantees is a fundamental requirement for their deployment in safety-critical settin

SafeCoEvo: Co-Evolving Safety Harnesses and Guards for LLM Agents at Test-Time

arXiv:2609.36580v1 Announce Type: new Abstract: LLM agents deployed in real-world environments continually encounter new tasks and safety risks, while execution feedback typically becomes available on

SAKI: Maximal-Coupling-Routed Teacher Supervision for On-Policy Distillation

arXiv:2609.36601v1 Announce Type: new Abstract: On-policy distillation (OPD) reduces train-test state mismatch by training a student on its own generated trajectories, but weak students may visit teac

Salt++: Context-Aligned Post-Training for Few-Step Streaming Multimodal Generation

arXiv:2609.36995v1 Announce Type: cross Abstract: Few-step streaming audio--video generation requires both causal modeling and step distillation, yet standard training recipes face two context-related

Sample Complexity of Equivariant Reinforcement Learning

arXiv:2609.36421v1 Announce Type: cross Abstract: Reinforcement learning (RL) is a powerful framework for robotic control, yet its practical application is often hindered by high sample complexity. Th

SCA: Spatial Credit Assignment for Reinforcement Learning of GUI Agents

arXiv:2609.36939v1 Announce Type: new Abstract: GUI agents automate tasks on digital devices by grounding language instructions in visual interfaces. Existing group-relative reinforcement learning imp